Chimpanzees appear to be unable to learn to speak. It is usual to attribute their lack of vocal skill to limitations of their vocal tract, and to the absence in their neocortex of any area corresponding to Broca's area in the human brain. The first signs of Broca's area in hominid endocasts are therefore taken to represent an evolutionary development of great significance. There are two outstanding questions. First, what exactly does Broca's area do? Secondly, why does Broca's area in one hemisphere play a much greater role in controlling speech than does the corresponding area in the other hemisphere? The following answers are proposed. (1) Broca's area seems to be concerned not with the production of individual sounds but with the regulation of sequences of sounds. Chimpanzees have no need for such an area because their natural calls are not made up by varying the sequential order of elementary units. (2) Cerebral dominance for speech may result from the fact that the vocal cords are innervated in the same way as other central organs, such as the tongue. Each hemisphere sends a projection, and the two projections overlap extensively so that either hemisphere can assume full control. It is argued that it is most efficient for a single hemisphere to dominate where a complex sequence of movements must be programmed. This reorganization has occurred for the production of song in some songbirds and for the control of the vocal cords in human speech.